The roar of F1 cars makes the world boil, and the "new" pulses that jump in the production line, workshop, and laboratory also make people's hearts surge. From today, follow Xiaojia into Jiading's innovative enterprises and feel the surging momentum brought by new quality productivity.
Stacked optoelectronics
Ultra-wide spectrum image sensor
If people rely on their eyes, nose, ears and other five senses to perceive the outside world, the intelligent era relies on sensors to achieve dynamic perception. With the rapid rise of smart cars and new energy vehicles, the transformation of the automotive industry is provoking the explosion of the sensor market. Among them, photoelectric sensors are widely used by converting photoacoustic, thermal, electronic, and biological processes into electrical signals.
There is such a company in the Shanghai Smart Sensor Industrial Park, which focuses on the research and development of multispectral stacked sensors to inject unlimited imagination into autonomous driving scenarios.
There are certain limitations of a single sensor, and the working mode of multiple sensors often faces technical bottlenecks, such as contradictions in time and information synchronization between sensors. The ultra-wide spectrum sensor developed by Diecheng Optoelectronics integrates the functions that previously required multiple chips to be completed on one chip, and covers the visible-infrared spectrum at the same time to achieve multispectral perception in the true sense.
Generally speaking, common cameras in life, such as mobile phone cameras, road monitoring, etc., are all visible light bands. Light sensors in the non-visible band, such as infrared night vision devices, are mostly used in the military and aerospace fields. The fusion of wide-spectrum perception into one sensor can greatly make up for the image recognition problems of visible light sensors in the case of direct sunlight, weak light at night and poor weather conditions, provide richer physical information for artificial intelligence machine vision, and improve the recognition rate and robustness of the algorithm (refers to the ability of the system, algorithm or method to maintain good performance in the face of abnormal situations or input changes). Just as camera imaging has been upgraded from black and white to color, ultra-wide spectral image sensors have ushered in the era of multispectral imaging.
Ultra-wide spectrum in-vehicle camera
Xiao Jia saw in the laboratory that in the scene of simulating weak visible light, thick smoke and fog and strong high beams at night, the infrared band can still detect targets well, especially when detecting people or other moving targets, and the imaging performance far exceeds that of existing visible light cameras. This technology can improve the reliability of autonomous driving perception in various complex roads, weather, and lighting conditions.
From left to right, color image, infrared image, ultra-wide spectrum image
"The new generation of ultra-wide spectral image sensor for the field of unmanned driving independently developed by Diecheng Optoelectronics uses chip stacking to achieve time-synchronous and space-synchronous ultra-wide-spectrum imaging on a chip, and has all-weather clear imaging capabilities in extreme scenarios such as dark night, backlight, smoke, dust, rain and fog. Zhong Youlin, deputy general manager of Diecheng Optoelectronics Technology, introduced that the current DC-A3 ultra-wide spectrum vehicle camera as an example can perceive color images and temperature distribution images at the same time, and cooperate with the newly developed algorithm to greatly improve the recognition accuracy. In addition, each pixel contains more information and has a higher recognition rate, which reduces the search space and requires less computing power when subsequent AI recognition is carried out.
Demonstration of a visual false target recognition scene
At present, Diecheng Optoelectronics has signed procurement contracts with customers such as autonomous driving companies in mining areas for unmanned vehicles in mining areas, and plans to carry out further in-depth cooperation with autonomous driving, assisted driving, security monitoring, rail transit industry monitoring, intelligent robots, special needs and other industries this year.
Source: Shanghai Jiading